Suppressive and pro-inflammatory roles for IL-4 in the pathogenesis of experimental drug-induced liver injury

Abstract

The pathogenesis of immune-mediated drug-induced liver injury (DILI) following halogenated anesthetics, carbamazepine or alcohol has not been fully elucidated. Detecting cytochrome P450 2E1 (CYP2E1) IgG4 auto-antibodies in anesthetic DILI patients suggests a role for IL-4 in this hapten-mediated process. We investigated IL-4-mediated mechanisms using our model of experimental DILI induced by immunizing BALB/c (WT) and IL-4^−/− (KO) mice with S100 liver proteins covalently modified by a trifluoroacetyl chloride (TFA) hapten formed following halogenated anesthetic metabolism by CYP2E1. WT mice developed more hepatitis, TFA and S100 antibodies (p<0.01), as well as T-cell proliferation to CYP2E1 and TFA (p<0.01) than KO mice. Additionally, WT CD4⁺ T cells adoptively transferred hepatitis to naïve Rag^−/− mice (p<0.01). Pro-inflammatory cytokines were expectedly decreased in TFA hapten-stimulated KO splenocyte supernatants (p<0.001); however, IL-2 and IFN-γ (p<0.05), as well as IL-6 and IL-10 (p<0.001) levels were elevated in CYP2E1-stimulated KO splenocyte supernatants, suggesting dual IL-4-mediated pro-inflammatory and regulatory responses. Anti-IL-10 administered to KO mice increased hepatitis, TFA and CYP2E1 antibodies in KO mice confirming a critical role for IL-4. This is the first demonstration of dual roles for IL-4 in the pathogenesis of immune-mediated DILI by suppressing auto-antigen-induced regulatory responses while promoting hapten-induced pro-inflammatory responses.

Introduction

Drug-induced liver injury (DILI) from idiosyncratic drug reactions account for approximately 13% of acute liver failure (ALF) cases in the United States 1–3, representing the third largest cause of ALF. ALF from acetaminophen or idiosyncratic drug reactions requires liver transplantation in 28% of patients and results in death in 50% of cases 2, 3. Evidence suggests that a type of DILI that follows the administration of halogenated volatile anesthetics, tienilic acid, dihydralizine, carbamazepine or alcohol is immune-mediated 4, 5. However, precise pathogenic mechanisms responsible for the initiation or subsequent development of hepatitis have not been fully elucidated.

Halogenated volatile anesthetics such as isoflurane, desflurane or halothane trigger immune responses to complex antigens formed from native hepatic proteins such as cytochrome P450 2E1 (CYP2E1) 6, 7 that have become covalently modified by a trifluoroacetyl chloride (TFA) hapten. This hapten is formed during the oxidative metabolism of the halogenated anesthetic by CYP2E1 8–10. Immune responses to these complex antigens occurring in secondary lymphoid organs generate cytokines that promote the development of inflammation in the liver 11. Additionally, TFA-haptenated native liver proteins induce both TFA antibodies, as well as auto-antibodies to CYP2E1 6, 12–14.

CYP2E1 auto-antibodies of the IgG4 subclass that are produced in the sera of persons with halogenated volatile anesthetic DILI form circulating, non-precipitating immune complexes that are believed to directly damage hepatocytes in these persons 15, 16. Detecting CYP2E1 auto-antibodies of the IgG4 subclass strongly suggests a role for IL-4 in the development of these auto-antibodies 17 and suggests that IL-4 promotes the development of immune-mediated DILI. Landmark studies clearly show that IL-4 and CD4⁺ T cells can induce B-cell proliferation and isotype switching from IgM to IgG4 and IgE 17, 18, which promotes the development of allergic and autoimmune diseases 19, 20. However, the mechanism of how IL-4 or any of the other cytokines initiates or regulates immune responses to auto-antigens or neo-antigens formed by drug haptens is not known. Additionally, although the formation neo-antigens has been strongly implicated in the initiation of immune-mediated DILI 6, 10–12, 14, 21, precise mechanisms for these processes have not been explored.

We have previously described a novel model of the initiation of experimental immune-mediated DILI induced by immunizing BALB/c mice with S100 proteins covalently modified with TFA haptens 11, 22. Hepatitis is evident at 3 wk whereas serum transaminase elevation occurs by 12 wk. The hepatic inflammatory infiltrate consists of neutrophils, eosinophils, T and B lymphocytes as well as NK and NKT cells 11.

Using this model of immune-mediated DILI, we have investigated the role of the CYP2E1 auto-antigen 6, 12, 14, the TFA hapten 8–10 and IL-4 in the initiation of immune-mediated DILI in susceptible BALB/c (WT) and IL-4-deficient (KO) mice. We verified that initiation of immune responses to CYP2E1 and TFA subsequently promote the development of anesthetic DILI by adoptively transferring hepatitis to female naïve Rag^−/− mice. We have found critical roles for CD4⁺ T-cell reactions to key antigens CYP2E1 and TFA in the initiation of anesthetic DILI by suppressing regulatory cytokine responses to the CYP2E1 auto-antigen and by inducing pro-inflammatory cytokine responses to TFA-haptenated proteins. This is the first demonstration of dual roles for IL-4 in the pathogenesis of autoimmune phenomena in immune-mediated DILI using a single animal model.

Results

Splenocytes from BALB/c (WT) proliferate in response to CYP2E1 and TFA

Previous studies demonstrated CYP2E1 auto-antibodies in anesthetic DILI patients, identifying CYP2E1 as the primary auto-antigen in anesthetic DILI 6, 12. Additional studies detected TFA antibodies in anesthetic DILI patients and strongly suggested that TFA was the primary hapten responsible for covalent modification of native proteins that trigger immune responses resulting in DILI 8–10. However, the lack of an animal model of immune-mediated, anesthetic DILI hampered studies investigating roles for CYP2E1 or TFA in the pathogenesis of anesthetic DILI. Hence, we first investigated the roles of CYP2E1 and TFA in the initiation phase of our model of immune-mediated, anesthetic DILI 11.

We measured proliferation to CYP2E1 and TFA using mixed splenocyte cultures that were collected from WT mice 2 wk following TFA-S100 immunizations (TFA-S100/CFA) and then re-stimulated with CYP2E1 or with TFA coupled to a non-self protein keyhole limpet hemocyanin (KLH-TFA) for 72 h. We found that proliferation to CYP2E1 and KLH-TFA was significantly higher in splenocytes from TFA-S100-immunized mice stimulated with 25 or 15 μg/mL of antigen (p<0.05, Fig. 1A and B) but not 10 or 5 μg/mL when compared with CFA-immunized mice. Splenocytes from WT mice did not demonstrate responses to KLH without the TFA hapten (results not shown).

Figure 1. Immune responses in WT splenocytes have a key role in experimental DILI. Stimulation indices were elevated in splenocyte cultures re-stimulated with the indicated concentrations of either (A) CYP2E1 or (B) KLH-TFA after isolation from TFA-S100 when compared to CFA-immunized WT mice (^*p<0.05, Mann-Whitney U. Mean ± SEM of pooled data from individual mice. Experiments were performed in triplicate with n = 4 mice/group). (C) Hepatitis was evident in naïve female Rag^−/− mice 10 days after receiving, by adoptive transfer, 1 x 10⁷ TFA-S100-immunized WT mixed splenocytes or CD4⁺ T cells re-stimulated with CYP2E1 (RAG 2E1 or RAG CD4⁺ T cells 2E1) or with TFA (RAG TFA or RAG CD4⁺ T cells TFA) when compared to RAG CFA or RAG Media (^*p<0.05, ^**p<0.01, Mann-Whitney U. Experiments were performed in triplicate with n = 2 mice/group). (D) Changes in percentage of CD3⁺CD4⁺ T cells, CD3⁺CD8⁺ T cells, B220⁺ B cells, DX5⁺ NK cells and DX5⁺CD3⁺ NKT cells following stimulation with CYP2E1 or KLH-TFA. Shown are mean of pooled data from individual mice. Experiments were performed in duplicate with n = 4 mice/group. Cytokine levels in splenocyte supernatants from (E) CYP2E1-re-stimulated WT splenocytes and (F) KLH-TFA-stimulated WT splenocytes (^*p<0.05, ^**p<0.01, ^***p<0.001, Mann-Whitney U. Mean ± SE of pooled data from individual mice. Experiments were performed in duplicate with n = 5 mice/group).

CD4⁺ T cells from WT mice induce hepatitis in Rag^−/− mice by adoptive transfer

To confirm that T-cell proliferation was associated with the development of DILI from anesthetics, we adoptively transferred hepatitis to naïve Rag^−/− mice. We found that WT-mixed splenocytes or CD4⁺ T cells isolated 14 days after TFA-S100 immunizations and re-stimulated for 72 h with 10 μg/mL of either CYP2E1 or TFA could adoptively transfer hepatitis to Rag^−/− mice, which was histologically evident after 10 days, whereas mixed splenocytes stimulated with medium alone or from CFA-immunized mice could not (Fig. 1C). These results strongly suggested that immune responses to CYP2E1 and to TFA generated by CD4⁺ T cells and possibly even CD8⁺ T, or NK cells as well as cytokines may induce anesthetic DILI.

CYP2E1 and TFA induce CD3⁺CD4⁺ T-cell proliferation but reduce NK and NKT cells

Proliferation of T and B lymphocytes as well as NK and NKT cells upon exposure to self antigens has been strongly implicated in the development of several autoimmune diseases 23–26. To identify which cells were responsible for proliferation to the key antigens in experimental anesthetic DILI, we analyzed the percent composition of these cells in mixed splenocyte cultures 72 h following in vitro culture. We found that there were 30 and 20% increases in CD3⁺CD4⁺ T cells following stimulation CYP2E1 and TFA, respectively, when compared with splenocytes cultured in medium alone. Conversely, stimulation with CYP2E1 induced 24 and 18% decreases in DX5⁺ NK cells and DX5⁺CD3⁺ NK T cells, whereas KLH-TFA stimulation decreased these cells by 15 and 9%, respectively (Fig. 1D). These findings suggested that increases in CD4⁺ T cells as well as decreases in potentially protective NK or NKT cells induced by CYP2E1 and TFA may have a role in the initiation of anesthetic DILI.

When analyzing CD8⁺ T cells, we found that re-stimulation with CYP2E1 increased CD3⁺CD8⁺ T cells by 8%, whereas TFA stimulation decreased these cells by a similar amount when compared with medium alone (Fig. 1D). Analysis of B cells showed that CYP2E1 decreased their numbers by 20% whereas TFA increased the composition of B cells in splenocyte cultures by 10% (Fig. 1D). Because interpretation of the immune responses to CYP2E1 were less clear while antibody responses to TFA haptens had been well documented in both rodents and humans, these analyses suggested to us that the CYP2E1 auto-antigen may have had an important role in the initiation of CD8⁺ T-cell responses whereas the TFA hapten had a more important role in the initiation of antibody responses in experimental anesthetic DILI.

CYP2E1 and TFA induce Th1 and Th2 responses in TFA-S100-immunized WT mice

To explain adoptive transfer of hepatitis to T-cell-deficient Rag^−/− mice by mixed splenocytes as well as purified CD4⁺ T cells from WT mice immunized with TFA-S100 and re-stimulated for 72 h with CYP2E1 or TFA, we measured supernatant cytokines from these splenocyte cultures treated as previously described. Similar to our previous findings 11, cytokines associated with TFA-S100-immunization and re-stimulation by CYP2E1 demonstrated significantly increased levels of primarily Th1 but also Th2 cytokines (Fig. 1E) whereas re-stimulation with TFA induced significant increases in Th1 cytokines alone (Fig. 1F). Interestingly, supernatant levels of the Th2 cytokine IL-13 in TFA re-stimulated cultures approached, but did not reach statistical significance. These studies suggested that the generation of CD4⁺ T cells by CYP2E1 and TFA by CYP2E1, during the initiation phase of our model triggered significantly elevated levels of primarily Th1 cytokines in response to CYP2E1 and TFA; however, Th2 cytokines generated in response to CYP2E1 may also have had a role in the initiation of anesthetic DILI. Even so, neither the cell-type induced by CYP2E1 or TFA nor the cytokines generated by these cells completely clarified the role of cells and cytokines in the initiation of DILI.

TFA-S100 significantly increases splenic and hepatic levels of IL-4 in WT mice

We had previously described cytokine levels in the liver at 3 wk in WT mice immunized with CFA±TFA-S100 11; however, we could not confirm a role for a particular cytokine in this process. To determine the role for the auto-antigen, hapten or adjuvant in the pattern of cytokine expression after 3 wk we determined whether there were significant differences in cytokine expression in TFA-S100-immunized mice when compared with CFA and S100 controls.

We confirmed that hepatic inflammation was significantly higher in TFA-S100 when compared with CFA-immunized mice (Fig. 2A and B). We found significantly higher levels of IL-4 in mixed splenocyte culture supernatants isolated 1 wk following immunization with TFA-S100 and re-stimulation with CYP2E1 or TFA when compared with cultures in medium alone (Fig. 2C), whereas IL-4 levels in spleen culture supernatants were not different when comparing CFA or S100-immunized mice. These findings suggested to us that as early as 1 wk following the first immunization with our haptenated auto-antigen, IL-4 expression is triggered by the critical DILI auto-antigen, CYP2E1 as well as the critical DILI hapten TFA. We also found that livers from TFA-S100-immunized mice had significantly higher levels of IL-4, in addition to IL-1β, IL-6, IL-13 and the chemokine MIP-2 after 3 wk (Fig. 2D). Taken together with our studies in patients, the present investigations suggest a critical role for IL-4 in the initiation and pathogenesis of immune-mediated DILI. However, whether this constituted a protective or destructive role was not clear. So next we wanted to directly assess the role of IL-4 by using IL-4-deficient mice. Interestingly, we also detected higher hepatic IL-10 levels in CFA-immunized mice when compared with the other groups which could support our findings of diminished inflammation in CFA-immunized mice and further confirm the S100 model 22, as well as our own, as models of hepatic inflammation.

Figure 2. TFA-S100 significantly increases splenic and hepatic levels of IL-4 in WT mice. (A and B) WT mice demonstrated significantly more hepatitis three weeks following TFA-S100/CFA immunizations when compared to immunization with CFA alone. Shown are representative liver tissue sections stained with H&E (0.5 μm thick, magnification 64×). (C) IL-4 in culture supernatants of mixed splenocytes obtained one week following TFA-S100 immunization and re-stimulated with either CYP2E1 or TFA in comparison with non-stimulated splenocytes (media) (^*p<0.05, Mann-Whitney U. Mean ± SEM of pooled data from individual mice. Experiments were run in duplicate with n = 4 mice/group). (D) Cytokine expression in liver tissue supernatants from CFA and S100 controls as well as TFA-S100-immunized BALB/c mice at 3 weeks (^**p<0.01, ^***p<0.001, Mann-Whitney U. Mean ± SEM of pooled data from individual mice. Experiments were run in duplicate with n = 5 mice/group).

IL-4 has a critical role in the development of experimental anesthetic DILI

The detection of elevated IL-4 in the splenocyte culture supernatants as well as the liver 1 wk and 3 wk following TFA-S100 immunizations, respectively, strongly suggested a role for IL-4 in this process. To directly assess the role of IL-4 in the development of anesthetic DILI, we induced experimental DILI in WT and IL-4^−/− (KO) mice and evaluated hepatitis in histological sections, supernatant cytokines in the liver and spleen, as well as serum antibodies after 3 wk.

We found that hepatic inflammation/injury was higher in WT when compared with KO mice (3.3±0.3 versus 1.6±0.2, mean±SE, p<0.01, Fig. 3A–C). Hepatic inflammation/injury was not significantly different in WT or KO mice given CFA alone (results not shown). Serum analyses for anti-TFA hapten antibodies demonstrated decreased IgG1 (p<0.001), but not IgG2a subclass antibodies (Fig. 3D) in KO mice, showing that IL-4 increased serum TFA IgG1 antibodies and suggested that IL-4 had a role in the generation of B-cell responses to the TFA hapten. Analyses of serum for anti-S100 auto-antibodies showed significantly decreased IgG1 and IgG2a subclass auto-antibodies to S100 in KO mice (p<0.001, Fig. 3D) suggesting that IL-4 also increased B-cell responses to this antigen and may be responsible for the development of these auto-antibodies which had also been previously demonstrated in WT mice 11.

Figure 3. IL-4 has a critical role in the development of experimental anesthetic DILI. Representative H&E histological section from (A) WT and (B) IL-4^−/− (KO) mice demonstrating significantly more hepatitis in WT mice (0.5 μm thick, representative data, 64×magnification). (C) Composite injury scores in WT and KO mice (^**p<0.01, Mann-Whitney U. Experiments were performed in duplicate with n = 5 mice/group). (D) Sera analysis for TFA and S100 antibodies WT and KO mice (^***p<0.001, ANOVA with Tukey's post test. Mean ± SEM of pooled data from individual mice. Experiments were performed in duplicate with n = 5 mice/group). (E, F) Cytokine analysis in liver tissue supernatants of WT mice and KO mice (^*p<0.05, ^**p<0.01, ^***p<0.001, Mann-Whitney U. Mean±SEM of pooled data from individual mice. Experiments were performed in duplicate with n = 5 mice/group).

When analyzing liver cytokine levels following TFA-S100 immunizations, we found that KO mice had significantly lower IL-12 and IL-13 (p<0.001), IL-5 (p<0.01) as well as IL-1β and TNF-α (p<0.05) when compared with WT mice (Fig. 3E and F), suggesting that IL-4 may have increased hepatic inflammation by increasing hepatic pro-inflammatory as well as regulatory cytokines. Cytokine levels in spleen were not significantly different between groups at this 3-wk time point (results not shown). These results showed that IL-4 had a critical role in the development of anesthetic DILI; however, whether these immune responses inferred a role for IL-4 in the initiation of experimental DILI needed to be addressed 11.

IL-4 initiates proliferation to CYP2E1 and TFA

To uncover critical pathogenic mechanisms induced by IL-4, we assessed T-cell proliferation in splenocytes collected from WT and KO mice 2 wk after immunization with TFA-S100 and stimulated for 72 h with CYP2E1 or TFA. We found that proliferation to CYP2E1 and KLH-TFA was higher in WT mice at all doses (p<0.01, Fig. 4A and B). This result demonstrated that IL-4 had a critical role in the initiation of immune responses to the key antigens responsible for the development of anesthetic DILI. We also found that splenocyte supernatants from KO mice re-stimulated with the TFA hapten had lower pro-inflammatory (IL-1β, p<0.01; IL-2, TNF-α, p<0.05), IL-4 effector (IL-13, p<0.05) and acute phase cytokines (IL-6, p<0.05) when compared with WT mice (Fig. 4D), suggesting that IL-4 had a critical role in the secretion of these cytokines following recognition of TFA-haptenated proteins by splenocytes from TFA-S100-immunized WT mice. In sharp contrast, when we measured cytokines in splenocyte supernatants from KO mice re-stimulated with CYP2E1, we found higher levels of IL-2 (p<0.05), IL-6 (p<0.001), and regulatory cytokines IL-10 (p<0.001) and IFN-γ (p<0.05) (Fig. 4C). Taken together with the demonstration of significantly diminished proliferation to CYP2E1 during the initiation phase (Fig. 4A), this surprising finding suggested that IL-4 may promote the development of immune-mediated anesthetic DILI by suppressing IL-10 or IFN-γ regulatory responses to the CYP2E1 auto-antigen during the initiation of this process.

Figure 4. IL-4 promotes T cell priming in experimental anesthetic DILI. Proliferation in response to various doses of (A) CYP2E1 and (B) TFA of WT and KO splenocytes (^**p<0.01, ^***p<0.001, Mann-Whitney U. Mean ± SEM of pooled data from individual mice. Experiments were performed in triplicate with n = 5 mice/group). Cytokine analysis of supernatants from (C) CYP2E1-stimulated (D) TFA-stimulated WT and KO splenocytes (^*p<0.05, ^**p<0.01, ^***p<0.001, Mann-Whitney U. Mean ± SEM of pooled data from individual mice. Experiments were run in duplicate with n = 5 mice/group).

IL-4-induced proliferation to CYP2E1 does not suppress regulatory T cells

To investigate possible mechanisms for IL-10 production in our splenocyte cultures, we determined whether immune responses to the CYP2E1 auto-antigen were being down-regulated by IL-10-producing Foxp3⁺CD25⁺ regulatory T cells 27, induced (Tr1 or Th3) regulatory T cells 28 or ordinary T cells performing regulatory functions. Two weeks following TFA-S100 immunizations, we found that there were no significant differences in the percent of Foxp3⁺CD25⁺CD4⁺ T cells when comparing WT and KO mice (Fig. 5A), suggesting that decreases in naturally occurring or induced regulatory T cells during initiation of immune responses to CYP2E1 did not account for decreased proliferation to key antigens and the decreased amount of immune-mediated hepatitis in KO mice. In fact, amounts of these cells were somewhat decreased in KO mice, which may be expected because prior studies suggested a requirement for IL-4 in the maturation of antigen-specific Foxp3⁺CD25⁺CD4⁺ T cells 29.

Figure 5. IL-4 suppression of responses to CYP2E1 does not involve regulatory T cells. (A) Percentage of Foxp3⁺CD4⁺CD25⁺ T cells in WT and KO mice. (B) TGF-β levels in splenocytes isolated from WT and KO mice following re-stimulation for 72 h with 10 μg/mL of CYP2E1. Mean ± SEM of pooled data from individual mice. Experiments were performed in duplicate with n = 5 mice/group. (C) Representative flow cytometric analysis of intracellular IL-10 and IFN-γ levels in splenocytes isolated from TFA-S100-immunized KO mice and activated with PMA and ionomycin.

Instead, immune responses to the CYP2E1 auto-antigen by KO mice may have indicated the presence of IL-10-producing Tr1 cells 28 or recently described conventional IL-10-producing Th1 cells performing regulatory functions 30, 31. Because TGF-β expression is one of the hallmarks of Tr1 cells 27, we measured TGF-β levels in splenocyte supernatants. We found that TGF-β levels were not increased in supernatants from KO mice splenocyte cultures re-stimulated with CYP2E1 (Fig. 5B), suggesting that suppressed proliferation to CYP2E1 was unlikely due to Tr1 cells. TGF-β levels were also similar in KO mice splenocyte supernatants re-stimulated with KLH-TFA when compared with WT mice (results not shown). Next we determined whether re-stimulation of TFA-S100-primed splenocytes with CYP2E1 induced Foxp3⁻, IFN-γ and IL-10 double positive T cells. However, we were not able to detect Foxp3⁻ IFN-γ and IL-10 double positive cells following CYP2E1 stimulation or PMA and ionomycin, while IL-10 and IFN-γ singly positive cells were present (Fig. 5C).

Our results suggested that during the initiation of anesthetic DILI, the source of IL-10 and diminished proliferation to CYP2E1 in KO mouse TFA-S100-primed splenocytes was most likely T cells performing regulatory functions. Interestingly, we did not find significant numbers of NK or NKT cells in these same cultures (results not shown). Nonetheless, we needed to clarify the role of IL-10 in the initiation of experimental anesthetic DILI.

Blocking IL-10 increases experimental anesthetic DILI in KO mice

To confirm a role for IL-10 in regulating the development of anesthetic DILI, KO mice received immunizations with TFA-S100 as previously described as well as 250 μg of anti-IL-10 or 200 μL PBS on days 14, 16, 18 and 20. Twenty-one days after the initial immunizations, hepatitis was assessed by histology, and sera were assessed for anti-CYP2E1 auto-antibodies as well as anti-TFA antibodies.

Hepatitis was increased in KO mice treated with anti-IL-10 when compared with KO mice treated with PBS following TFA-S100 immunizations (p<0.05, Fig. 6A–C). Treatment of KO mice with anti-IL-10 following TFA-S100 immunizations also increased levels of anti-CYP2E1 IgG and TFA IgG1 antibodies when compared with PBS/TFA-S100-treated mice (p<0.05, Fig. 6D and E). This finding was very important because we had previously demonstrated that anti-TFA IgG1 antibodies correlate with hepatitis in this model 11. Moreover anti-CYP2E1 auto-antibodies had been detected in anesthetic DILI patients 14, 15.

Figure 6. Anti-IL-10 increases hepatitis, as well as CYP2E1 and TFA antibodies in KO mice. TFA-S100 immunization in combination with injection of (A) 200 μL PBS or (B) 250 μg anti-IL-10 blocking antibodies on days 14, 16, 18 and 20. Representative data are shown (64× magnification). (C) Histology scores, (D) sera anti-CYP2E1 IgG autoantibodies and (E) anti-TFA IgG1 antibodies were analyzed for anti-IL-10- vs. PBS-treated female KO mice (mean ± SE, ^*p<0.05, Mann-Whitney U. Pooled data from individual mice. Experiments were performed in duplicate with n = 4 mice/group).

Discussion

We have shown that analogous to clinical findings, IL-4-driven immune responses to CYP2E1 and TFA have critical roles in the initiation and severity of experimental anesthetic DILI. Surprisingly, re-stimulation of IL-4-deficient mouse (KO) splenocytes with CYP2E1 induces significantly elevated levels of IL-2, IL-6, IL-10 and IFN-γ, suggesting that IL-10 – or even IFN-γ-producing T cells may have a role in down-regulating T-cell proliferation to CYP2E1. Blocking IL-10 in KO mice increased hepatitis as well as CYP2E1 and TFA antibodies in KO mice confirming a regulatory role for IL-10 in the development of experimental anesthetic DILI. Our studies strongly suggest dual roles for IL-4 in the development of immune-mediated anesthetic DILI by down-regulating protective responses to the CYP2E1 auto-antigen while promoting pro-inflammatory responses to TFA.

Previous clinical investigations have determined that immune responses to CYP2E1 and TFA have critical roles in the development of anesthetic DILI. These elegant studies clearly demonstrated antibodies to CYP2E1 and TFA in the serum of persons who developed anesthetic DILI or Id-ALF 6, 9, 12. Because a hapten-based immune-mediated mouse model of DILI is a recent development 11, we can now determine the exact roles for these key proteins and directly investigate the pathophysiology responsible for the development of immune-mediated DILI.

No previous study has identified critical cytokines associated with the initiation of anesthetic DILI, in mice or patients. However, elevated acute phase cytokines have been demonstrated in toxicity models following the administration of intra-peritoneal halothane to guinea pigs 32 and large doses of intra-peritoneal halothane to mice 33, suggesting that these studies represent important models of anesthetic Id-ALF from direct toxicity by halothane. We believe that our immune-mediated approach may parallel what occurs during the clinically quiescent phase of anesthetic DILI in patients, where hepatitis is undetectable because of normal or mildly elevated serum transaminase levels, which are late findings. Demonstrating immune responses to key antigens associated with anesthetic DILI in patients (Fig. 1) as well as a critical role for IL-4 (Figs. 2–4), which has been clearly implicated by the demonstration of IgG4 auto-antibodies during active anesthetic DILI 15, 16 firmly supports the translational significance of this model.

We have demonstrated significantly elevated TNF-α and IL-1β, as well as IL-6, IL-10, IL-13, and IFN-γ in the splenic priming phase of our model prior to the development of significant hepatitis. TNF-α and IL-1β have significant roles in the development of DILI following acetaminophen 34. Moreover, these same cytokines have been demonstrated in patients with infectious hepatitis and in autoimmune hepatitis models 11. Our studies agree with these previous findings and suggest that toxic and autoimmune responses, as well as Th1, Th2 and regulatory cytokines have roles in the initiation phase of immune-mediated DILI. More importantly our studies directly correlate the production of these cytokines with key antigens that have been strongly associated with the development of anesthetic DILI 6, 9, 10, 12, 14, 15.

Detecting increased IL-4 in splenocyte supernatants and subsequently in the liver of TFA-S100-immunized mice when compared with S100 and CFA-immunized control mice clearly indicates a critical role for IL-4 in a hapten-mediated inflammatory process in the liver (Fig. 2C and D). Previous studies have correlated IL-4 with the development of immune-mediated T-cell hepatitis 35 where IL-4 in conjunction with CD4⁺ T cells induces liver injury by activating the Jak1/STAT6 pathway in hepatocytes and NKT cells resulting in the production of eotaxin and IL-5 36. Similar to these studies, we have found that IL-4 is required for the initiation of anesthetic DILI. We also show that KO mice develop significantly less hepatitis as well as hepatic cytokine responses (Fig. 3). However, we did not find significant differences in hepatic eotaxin when comparing TFA-S100-immunized WT and KO mice (results not shown). Additionally, we demonstrate significant decreases in splenic NK and NKT cells presumably triggered by CYP2E1 and TFA. Thus our studies show a direct effect of immune responses to these antigens on the survival of potentially protective NK and NKT cells, and further suggest that additional regulatory pathways may influence the development of immune-mediated DILI from anesthetics or other drugs that use similar mechanisms. Along these lines, IL-4 can increase mRNA and protein expression of CYP2E1 in hepatocytes 37, 38. In this way, IL-4 may have a similar role in the development of halogenated volatile anesthetic DILI in susceptible persons by increasing metabolism of halogenated volatile anesthetics, thus creating more reactive haptens and subsequently more neo-antigens that could initiate anesthetic DILI.

Surprisingly we have found increased levels of IL-2, IL-6 IL-10 and IFN-γ in splenocytes from TFA-S100-immunized KO mice stimulated with CYP2E1 (Fig. 4C). This finding suggests to us that IL-4 may have an additional role in the suppression of potentially protective responses by IL-10-producing T cells propagated by pathogenic T cells that secrete IL-2, similar to previously described Th1 cells performing regulatory functions 30. Although we did not find significantly increased antigen-specific Foxp3⁺CD25⁺CD4⁺ regulatory T cells, this may have been affected by the deficiency of IL-4 in our KO mouse, because IL-4 is required for the maturation of these cells 29. Although our findings clearly suggest that IL-4 has a role in the development of anesthetic DILI in our model and in patients, it is clear that some KO mice can develop reasonable amounts of hepatitis. This finding could suggest that the primary role for IL-4 is in the initiation phase whereas an alternative role later in disease may be regulated by additional immunomodulatory mechanisms induced by CYP2E1, IL-10 or even IFN-γ. Another possibility is that suppression of CYP2E1 immune-mediated responses in the initiation of this model by IL-10 may promote the development of chronic, but decreased disease in KO mice, a mechanism recently suggested in the development of chronic cutaneous leishmaniasis 31.

We cannot ignore our detection of IL-4 in the liver of KO mice (Fig. 3E) and wonder whether or not IL-4 contributed to the development of hepatitis in some KO mice (Fig. 3C). Expectedly, IL-4 levels are below levels of detection in the spleen. We have contacted Jackson Laboratories (Bar Harbor, Maine), the source of our mice and confirmed our mice to be IL-4-deficient. Even so, these experiments have been repeated several times by us with similar results. We have concluded that these levels may not be checked by others and the IL-4 deficiency status may have been assumed in all organs. We elected to keep the IL-4 levels as we detected them in the Results section, because this is truly what we found.

Our studies suggest that in immune-mediated DILI from halogenated volatile anesthetics or other drugs that generate drug haptens, CD4⁺ and possibly CD8⁺ T-cell responses to key antigens such as CYP2E1 and TFA are primarily responsible for the initiation of immune responses that result in hepatitis. Although CD4⁺ T cells generated during the initiation of anesthetic DILI trigger both Th2 and Th1 cytokines, IL-4 has a key role in the initiation of this process. In the absence of IL-4, IL-10-producing T cells diminish immune responses to CYP2E1, suggesting that IL-4 may directly or indirectly decrease the activation or survival of these T cells. Blocking IL-10 in KO mice increases hepatitis as well as anti-CYP2E1 auto-antibodies and anti-TFA antibodies, confirming a critical role for IL-10 in these processes (Fig. 7). This is the first demonstration for dual roles for IL-4 in experimental immune-mediated DILI that may partially explain the development of autoimmune symptoms in patients with anesthetic DILI 15 or immune-mediated DILI from other drugs that use similar mechanisms. Ongoing investigations will further define roles of IL-4, neo-antigens, additional immune cells and other cytokines such as IFN-γ in immune-mediated DILI from halogenated volatile anesthetics or other drugs using similar mechanisms such as, tienilic acid, dihydrazaline, carbamazepine and alcohol.

Figure 7. Proposed mechanism of immune-mediated DILI. In our model of experimental (murine), immune-mediated, anesthetic DILI, BALB/c mice develop an IL-4-promoted hypersensitive response to the TFA hapten and S100 liver proteins resulting via adjuvant–induced splenocyte TNF-α, IL-1β, IL-6 production. IL-4-dependent splenocyte proliferation is also demonstrated in response to human CYP2E1 and TFA through suppression of IL-10 and possibly IFN-γ regulatory responses. Immune responses to CYP2E1, S100 and TFA also activate mast cells that recruit neutrophils to the liver. Directly or indirectly through IL-10, a predominantly IL-4-driven Th2-type response leads to S100 and CYP2E1 auto-antibody production as well as TFA hapten antibodies.

Materials and methods

Materials

ACK Lysing Buffer and PBS were from Quality Biological (Gaithersburg, MD); alkaline phosphatase-conjugated rabbit anti-mouse IgG1 and IgG2a and IgG2b and Golgi Stop™ from BD Biosciences (Woburn, MA); alkaline phosphatase substrate kit from Bio Rad; mouse cytokine/chemokine 22-plex pre-mixed LINCOplex kit from Millipore (St. Charles, MO); Difco Bacto Adjuvant Complete Freund H37 Ra (CFA) from Fisher Scientific (Pittsburgh, PA); FBS, L-glutamine, HEPES Buffer Solution (1 M), MEM penicillin-streptomycin, RPMI 1640 and Trypan Blue Stain, from Invitrogen™, (Carlsbad, CA); goat anti-mouse IgG (heavy and light chains) alkaline phosphatase conjugate (AKP, Chemicon International, Temecula, CA); Gill's Number 1 hematoxylin, from Sigma-Aldrich (St. Louis, MO); human CYP2E1 from Gentest, BD Biosciences; Immulon 2HB^® microtiter 96-well plates, from ISC BioExpress (Kaysville, UT); ketamine and xylazine, from, from Penn Veterinary Supply (Lancaster, PA); methyl ³H thymidine, from Perkin-Elimer (Boston, MA); PE anti-mouse/rat Foxp3 Staining Set, from eBioscience (San Diego, CA); pertussis toxin, from List Biologicals (Campbell, CA); rat anti-mouse IL-10 monoclonal antibody (clone JESS-2A5), Rat IgG1 Isotype control, and Quantikine cytokine ELISA kits, from R&D Systems (Minneapolis, MN).

Mice

Eight to ten-wk-old, female, inbred BALB/c (WT), IL-4^−/− (KO) mice on BALB/c background as well as Rag^−/− mice on BALB/c background, purchased from the Jackson Laboratory were maintained under pathogen-free conditions at our animal facility. The IL-4-deficient mice were derived from targeted deletion of IL-4 performed in a BALB/cJ-derived ES cell line. Targeted ES cells were then injected into BALB/c blastocysts. The colony has been maintained by brother–sister mating of offspring from these blastocysts. This colony was derived solely on a BALB/cJ background. The Rag1^tm1Mom mutant strain was developed by Dr. Peter Mombaerts in the laboratory of Dr. Susumu Tonegawa at the Center for Cancer Research, Massachusetts Institute of Technology. A replacement targeting vector with the Pgk-neo marker was used. Homologous recombination of the targeting vector resulted in a 1356-bp deletion in the 5′ end of the coding sequence. The 129S7/SvEvBrd AB1 ES cell line was used. The BALB/c congenic strain was generated by Dr. Bob Coffman by backcrossing mice carrying the Rag1^tm1Mom mutation seven times to BALB/cAnNTac inbred mice. The approximate control is the BALB/cJ mouse. The strain is maintained by homozygous sibling matings. Approval for all procedures was obtained from the Animal Care and Use Committee of the Johns Hopkins University.

Induction of hepatitis with TFA-haptenated cytosolic S100 (TFA-S100)

To induce hepatitis, WT or KO mice were immunized s.c. with two doses of 200 μg of syngenic S100 covalently altered by TFA on days 0 and 7, emulsified in an equal volume of CFA at the base of the neck, and 500 ng of Pertussis toxin i.m. in the right hind flank on day 0 as previously described 11.

Anti-IL10 administration

Experimental anesthetic DILI was induced in KO mice as described 11. In addition, mice received 250 μg/200 μL PBS anti-IL-10, isotype control or 200 μL PBS on days 14, 16, 18 and 20. The mice were killed 3 wk after the initial immunization.

Proliferation assays

One or two weeks after the initial TFA-S100 immunization, whole splenocytes were isolated and 2.5×10⁵ cells/100 μL PBS/2% FBS were loaded into 96-well plates with 100 μL aliquots of Media, Con A, 25, 15, 10 or 5 μg/mL CYP2E1, trifluoroacetylated KLH (KLH-TFA) or KLH alone. Plates were incubated for 72 h at 37°C, 5% CO₂, 95% air (humidified), pulsed with 10 μL, 1 μCi/well methyl ³H thymidine and then stored at −20°C. Experiments were performed at least in duplicate with N=4 mice/group.

FACS analysis of the cellular composition of splenocyte culture supernatants

Splenocytes were isolated 2 wk following initial TFA-S100 immunization, re-stimulated with media±CYP2E1 or KLH-TFA (10 μg/mL), incubated for 72 h at 37°C, 5% CO₂, 95% (humidified), pooled by stimulation antigen, counted, washed with PBS/2% FBS, and stained with the antibody combinations described in the next paragraph. For intracellular cytokine staining, cells were treated with Golgi-Stop™ at 24 h, harvested by 48 h, pooled by antigen stimulation, permeabilized and fixed. (Experiments were run in duplicate with N=4 mice/group).

We utilized 1:100 dilutions of the following FITC/PE/CyChrome™ or APC-labeled antibody combinations: CD4 (L3T4, clone H129.19)/CD8a (Ly-2, clone 53.7.7)/CD3e (clone 145-2C11); CD49b Pan-NK Cells (clone DX5)/CD45R B220 (clone RA3-6B2)/CD3e (clone 145-2C11); CD11c (clone HL3)/CD8a (Ly-2, clone 53.7.7)/CD3e (clone 145-2C11); F4/80 (clone BM8)/Ly-6G Gr-1 (clone RB6-8C5)/CD45 (clone 30-F11) and CD4 (L3T4, clone H129.19)/Foxp3 (clone FJK-16s)/CD25 (clone PC61). For intracellular cytokine staining, CD4⁺ T cells were stained with 1:100 dilutions of the following FITC/PE, CyChrome™ antibody: IFN-γ (clone XMG1.2)/IL-10 (clone JES5-16E3)/CD4⁺ (G3T4, clone H129.19). All antibodies were from BD Biosciences Pharmingen (San Diego, CA) except F4/80 and Foxp3, which were from eBiosciences. Cells were fixed with PBS/4% paraformaldeyde and analyzed by flow cytometry within 3 days.

Adoptive transfer experiments

Whole splenocytes were isolated from CFA or TFA-S100-immunized mice as described above and cultured for 72 h in media±CYP2E1 or KLH-TFA (10 μg/mL). The CD4⁺ T cells used for adoptive transfer experiments were obtained using magnetic beads MACS^®Cell Separation System (Miltenyi Biotech, Auburn, CA) as per manufacturers instructions on whole splenocytes from TFA-S100-immunized BALB/c mice cultured in media or re-stimulated with either CYP2E1 or TFA (10 μg/mL). Prior to adoptive transfer, splenocytes or CD4⁺ T cells/mL isolated from in vitro whole splenocyte cultures were washed several times in PBS and viable cells counted. 1×10⁷ cells/mL were injected into separate Rag^−/− mice using the tail vein. Ten to fourteen days later, the mice were killed and livers assessed for hepatitis. Experiments were done in triplicate with N=2 mice/group.

Histological analysis

Liver tissue sections (0.5 μm thick) were fixed in 10% neutral buffered formalin and stained with H&E and scored for inflammation and injury using a previously described grading system 11: Grade 0, no portal tract, lobular inflammation or hepatocyte necrosis; Grade 1, minor periportal or lobular inflammation without necrosis; Grade 2, periportal or lobular inflammation involving <50% of the liver section; Grade 3, periportal or lobular inflammation involving ≥50% of the liver section; Grade 4, inflammation with bridging necrosis. Experiments were performed in duplicate with N=5 mice/group.

ELISA to detect TFA or S100 antibodies

Blood collected on day 21 following the initial immunization was allowed to clot at room temperature, centrifuged at 2000 rpm for 15 min at 4°C, serum was removed and antibodies were detected by ELISA as previously described 11 using 0.5 μg/100 μL BALB/c S100, TFA bound to ovalbumin or CYP2E1. Secondary antibodies consisted of goat anti-mouse IgG (heavy and light chains) alkaline phosphatase conjugate (AKP), mouse IgG1 AKP or mouse IgG2a AKP in PBS/2% FBS (1:1000). The OD (405 nm) of TFA auto-antibodies was determined after a 30-min incubation with substrate, whereas S100 and CYP2E1 auto-antibody levels were detected after 60 min. Experiments were performed in duplicate with N=5 mice/group.

Cytokine measurements

Liver and spleen samples from individual mice were snap-frozen and stored at −80°C until homogenized using 10% weight/volume in 2% MEM 11. Cytokine levels in individual organs were standardized by converting cytokine levels to pg/g of tissue. Splenocyte culture supernatants were collected and stored at −80°C until tested. Cytokines were measured using Quantikine cytokine ELISA kits purchased from R&D Systems. Experiments were performed in duplicate with N=5 mice per group.

Statistical analysis

Statistical analyses were performed using GraphPad^® Prism Version 3.02 for Windows (GraphPad^® Software, San Diego, CA). Histology scores, proliferation assays and cytokines were analyzed using Mann–Whitney U-test. Antibodies were analyzed by ANOVA with Tukey's post-test. A p value <0.05 was considered significant. ^*p<0.05; ^**p<0.01; ^***p<0.001.

Acknowledgements

D.B.N. is supported by NIHR21DK075828. This work is supported in part by a grant from the American Autoimmune Related Diseases Association, Mr. and Mrs. Joseph Scoby and The Gail I Zuckerman Foundation.

Conflict of interest: The authors declare no financial or commercial conflict of interest.